Why committing to volume can lower a rate, when it does not, and how to size a commitment you can still meet in a bad quarter.
Two offers land on your desk for the same rack space, and the only real difference between them is a single clause: one asks you to commit to a minimum volume over the term, the other lets you scale up or down each month. The committed offer carries a lower rate. Before you sign either one, understand exactly what that lower rate is paying for, and what it costs you if a quarter goes badly.
What a take-or-pay commitment commits you to
A take-or-pay clause sets a minimum volume, usually expressed as committed load or as energised megawatts, that you agree to pay for whether or not your fleet draws it. If your hosted capacity runs below the committed floor in a given period, you still pay as if it had run at the floor. The clause is not unusual, and it is not automatically unfavourable. It is a trade: you give the operator a predictable revenue floor, and the operator gives you a rate that reflects that certainty rather than the higher rate that comes with month-to-month flexibility.
Read the clause for what it measures. Some commitments are set against contracted capacity regardless of uptime, which means an outage on the operator's side still counts towards your obligation unless the contract carves that out explicitly. Others are set against power actually delivered, measured at the meter, which shifts more of the operational risk back onto the operator. The two versions look similar on the page and behave very differently in a bad month, so ask which one you are being offered before you compare the headline rate against a flexible alternative.
Why a fixed commitment can lower the rate
The discount is not a sales gesture. It reflects something real on the operator's side: a known revenue floor changes how power gets procured, how staff time gets planned, and how a site's capacity gets allocated among customers competing for the same rack space. An operator taking on a fleet with a firm volume commitment can plan around that number with more confidence than around a customer who might scale down without notice, and that certainty is worth something in the rate. This is the same logic that sits behind a minimum term, covered in more detail in our guide to the whole hosting agreement. Commitment on one side of the table generally buys a better number on the other side, and the two clauses are best read together rather than in isolation.
Sizing a commitment you can survive
The number that matters is not what your fleet can draw at full output. It is what your fleet is likely to draw in a genuinely bad quarter: a hardware batch offline for repair, a curtailment event, a period where part of the fleet is being redeployed or sold. A commitment sized against your best quarter looks attractive on the page and turns into a shortfall payment the first time reality falls short of that number. Work backwards from your worst realistic month over the term, not your average one, and size the floor against that. If you cannot say with confidence what your fleet's output looked like in its worst month over the past year, establish that before you commit to a number for the next one.
The measurement period matters as well. A clause that averages your draw across the whole term is far more forgiving than one that checks compliance every month, since the first absorbs a bad month against a good one and the second does not.
Common exceptions and what falls outside them
Most take-or-pay clauses carve out specific circumstances where a shortfall does not count against you. A period where the operator has failed to deliver energised capacity is a common one. Scheduled maintenance, disclosed in advance, is another. A grid curtailment event outside either party's control is sometimes included and sometimes not, so ask directly rather than assuming. What is rarely carved out is ordinary hardware attrition on your side: a batch failing, a unit being sold, a rack sitting idle while you source a replacement. That risk generally stays with you, which is exactly why the sizing question above matters more than the exceptions list does.
What to ask before you sign
A short set of questions turns a take-or-pay clause from a number on a page into something you can plan against.
- Is the commitment measured against contracted capacity, or against power actually delivered?
- Is compliance checked monthly, or averaged across the whole term?
- What specific events are excluded from the shortfall calculation, and is downtime caused by the operator one of them?
- What is owed if you fall short: a straight shortfall payment, a true-up against future usage, or something else?
- Can the committed volume be adjusted during the term, and what is the process for that?
Asking them does not signal doubt. An operator offering a committed rate should be able to answer each one plainly, and how clearly it answers tells you as much about the relationship as the rate itself does.
The same logic behind a volume commitment also shows up in how a contract handles the end of the relationship. Our piece on minimum terms and notice periods looks at that side of the same trade: what you get for locking in, and what it costs to get out. Our minimum deployment is 250 kW, and the terms that apply are set out in the service agreement, which comes with a quote from our hosting rate page.
